Main

Hominoids underwent a major evolutionary radiation during the Miocene epoch, with over 40 widely recognized species in at least 30 genera1. Despite this multitude of taxa, only about one-third are known from any cranial remains, and no more than half a dozen preserve any substantial portion beyond the face and palate2. Thus, much about hominoid cranial evolution remains poorly understood, especially with respect to the ancestral morphology that gave rise to the clade containing extant apes and humans. Importantly, the African fossil record lacks any reasonably complete hominoid crania between 17 and 7 million years (Myr) ago, and no cranial specimens are known at all from between 14 and 10 Myr (refs 3, 4, 5, 6), greatly hampering the analysis of hominoid evolution. The KNM-NP 59050 cranium reported here was recovered from Napudet (South Turkana, Kenya) and dated to 13 Myr; it thus falls within this critical, yet poorly represented, period. The infant specimen is nearly complete, but is missing the deciduous dental crowns (Fig. 1a–d and Extended Data Fig. 1a–f). The unerupted adult dentition, brain endocast, and bony labyrinths were visualized using propagation phase-contrast X-ray synchrotron microtomography (PPC-SR-μCT; Fig. 1e–h)7. The crown morphology of the fully formed I1s and M1s, as well as the partly formed M2s (Fig. 2 and Supplementary Data 1), indicate that the specimen warrants attribution to a new species in the genus Nyanzapithecus.

Figure 1: KNM-NP 59050.
figure 1

a–d, Specimen as preserved in anterior view (a), superior view (b), inferior view (c), and left lateral view (d). e–h, Three-dimensional visualizations based on X-ray microtomography, in views matching a–d, and with the bone rendered transparent to show the deciduous dental roots (beige), the unerupted permanent tooth crowns (grey), the bony labyrinths (green), and the endocast (blue transparent in e–g and beige inh; the olfactory fossa marked by the blue line placed directly underneath). Scale bar, 5 cm.

PowerPoint slide

Full size image
Figure 2: Unerupted permanent dentition.
figure 2

a–g, Three-dimensional X-ray microtomography-based visualization of the left I1 to M2, respectively, shown from left to right in occlusal, mesial, lingual, distal, buccal/labial views. h–n, The right I1 to M2 as shown for the left side. In occlusal view, the lingual side of the crown is down. Scale bar, 5 mm.

PowerPoint slide

Full size image

Systematic palaeontology

Order Primates Linneaus, 1758

Suborder Anthropoidea Mivart, 1864

Infraorder Catarrhini Geoffroy, 1812

Superfamily Hominoidea Gray, 1825

Subfamily Nyanzapithecinae Harrison, 2002

Genus Nyanzapithecus Harrison, 1986

Nyanzapithecus alesi sp. nov.

Etymology. Specific name taken from the Turkana word for ancestor, Ales.

Holotype. KNM-NP 59050, an almost complete infant cranium preserving fully formed but unerupted I1 and M1 crowns, as well as partly formed crowns of all other permanent teeth, except the not yet initiated M3s.

Locality and horizon. Napudet (2° 57′ N, 35° 52′ E), Turkana Basin, Kenya, Emunyan Beds, Brown Bedded Tuffs (Extended Data Fig. 2a).

Geological age. 13 Myr.

Diagnosis. A large species ofNyanzapithecus, with M1 significantly larger than in N. pickfordi (P < 0.05),N. harrisoni (P < 0.01), and probably N. vancouveringorum (Fig. 3a and Extended Data Table 1a; one-tailedt-test, Bonferroni corrected). The upper molars of N. alesi differ from those of N. vancouveringorum in being more waisted, and in having higher and more inflated molar cusps, a very restricted trigon, and a mesial shelf.N. alesi differs from N. vancouveringorum and N. harrisoni in having an M1 with a paracone approximately the same size as the metacone, and a protocone much larger than the hypocone. A reduced lingual cingulum also distinguishes N. alesi from N. harrisoni, but not from either N. vancoveringorum or N. pickfordi. N. alesi further differs from N. pickfordi in that the prehypocrista of the M1 meets the base of the protocone rather than the crista obliqua.

Figure 3: Dental metric comparisons of KNM-NP 59050.
figure 3

a, M1 area (maximum mesiodistal × maximum buccolingual) compared with that of published Nyanzapithecus species.b, M1 shape (maximum mesiodistal/maximum buccolingual, MD/BL) compared with extant and fossil hominoids. KNM-NP 59050 (dashed line) falls exclusively within the nyanzapithecine range among fossils. c, Relative I1 size (I1 maximum mesiodistal/M1 maximum mesiodistal) compared with extant and fossil hominoids. Nyanzapithecus, including KNM-NP 59050, falls closest toSymphalangus. Samples are given in Supplementary Data 2. For each sample the mean, the range between the first and third quartiles (box), and the highest and lowest values (whiskers) are indicated, with small ticks marking measured values in the sample.

PowerPoint slide

Full size image

Cranial morphology

KNM-NP 59050 is a nearly complete but somewhat distorted cranium of an infant primate (Fig. 1). The cranium is slightly crushed bilaterally and the posterior portion of the basicranium is both broken and distorted. All the deciduous tooth crowns are broken off, but their roots are preserved. The permanent teeth are unerupted, with the right I1 being visible in its crypt.

The overall dimensions of KNM-NP 59050 are similar to those ofSymphalangus crania of equivalent dental age, except for the maxillo-alveolar size, which is simiar to Hoolock (Extended Data Table 1b). Relative to overall cranial size, the snout is small as in juvenile hylobatids, and smaller than in extant juvenile hominids (Extended Data Figs 3 and 4a–c). This difference between hylobatids and hominids persists into adulthood (Extended Data Fig. 4b, d), and assuming that N. alesi followed the same pattern, its snout would have been relatively small as an adult, unlike that ofAfropithecus and Saadanius.

The orbits appear large, but are well within the expected range for an extant juvenile hominoid of its size (Extended Data Fig. 4e). The orbits are slightly taller than wide, which may reflect the bilateral distortion (Extended Data Fig. 3a, b). The supraorbital costae/ridges are poorly defined, similar to other infants, and in contrast to the condition observed in most adult extant hominoids. The lateral and inferior margins of the orbit are protruding and bar-like, which is unlike the flat/continuous margin seen among infant and adult great apes but similar to the morphology observed in hylobatids (infants and adults), pliopithecoids, and, to a lesser extent, Micropithecus8. The consistency of this feature among extant adult and juvenile apes suggests that it is ontogenetically stable. The interorbital area is relatively wide compared with the condition in extant juvenile hominids, but near the mean width for juvenile hylobatids, again a pattern probably maintained into adulthood and seen in Afropithecus andTurkanapithecus (Extended Data Fig. 5a, b). The lacrimal fossa is in line with the medial orbital margin, as in extant hylobatids and Victoriapithecus, rather than being clearly within the orbit as inPan and Gorilla, or clearly anterior to the orbital margin as commonly seen inAegyptopithecus9. Nasion is located about a third of the way down between the levels of the inferior and superior orbital margins. Although the edge of the nasal aperture is broken, it is likely that the premaxillae ended at the lower edge of the nasals, as in other Miocene hominoids10, but unlike the condition seen in most cercopithecoids, juvenile or adult, or Aegyptopithecus, where the premaxillae extend superiorly between the nasals and the maxillae. The nasals are tall and rectangular; they do not appear to be hour-glass or diamond-shaped, as in chimpanzees of all ages, or to broaden inferiorly as in some cercopithecoids and gorillas (infants and adults). The nasal aperture is relatively narrow, as in Pongo (Extended Data Fig. 5c, d), and close to the width of the premaxillae.

The malar region is oriented posteroinferiorly, as in adult hylobatids and many extant juvenile catarrhines, contrasting with a more vertical or anteroinferior orientation in most adult Old World anthropoids, including pliopithecoids, Victoriapithecus, Ekembo, and Turkanapithecus. In extant catarrhine taxa, malar orientation changes during ontogeny, becoming more anteroinferior by adulthood. This makes it difficult to predict what the adult condition in KNM-NP 59050 would have looked like, but the preserved morphology is most similar to adult and juvenile hylobatids. The root of the zygomatic arch originates close to the alveolar margin, which is similar to the condition seen in Aegyptopithecus and Rangwapithecus, and probably reflects a primitive retention11. Other Miocene catarrhines, as well as hylobatids of all ages, vary in this feature. However, the inferior base of the zygomatic root is positioned relatively higher on the maxilla of juvenile and adult extant great apes, a notable distinction from many other catarrhines.

The greatest cranial breadth is between the well-developed supramastoid crests, unlike in hylobatids of the same dental age, where the parietal bosses still project more laterally, and the crests are incipient (Extended Data Fig. 3b, c). The coronal suture appears to be oriented in a mediolateral direction, as in extant hominids, Ekembo, and Turkanapithecus. This contrasts with the anteriorly oriented, V-shape coronal suture seen in many cercopithecoids, platyrrhines, and hylobatids (Extended Data Fig. 3). Related to this configuration, the frontal is anteroposteriorly shorter than the parietal in the midsagittal plane (Extended Data Fig. 5e, f). The squamous portion of the left temporal is nearly a hemi-circle, with the highest point of the temporoparietal suture near the anteroposterior middle of the bone, above the posterior root of the zygomatic arch, the glenoid fossa, and the external acoustic meatus. This condition is most similar to that seen inSymphalangus infants and adults, but it is different from the condition observed in other hylobatids and great apes, in which the suture is oriented horizontally or anteroinferiorly, respectively.

The external acoustic meatus is represented by a completely ossified tubular ectotympanic, unlike pliopithecoids and Pliobates, in which the inferior surface is not fully ossified12. It is fused anteriorly to a prominent postglenoid process. The posterolateral orientation of the meatus is most similar to the condition seen in some great apes and cercopithecoids (infants and adults), but is different from that observed among hylobatids, where the ectotympanic is typically oriented more anterolaterally. The lateral pterygoid plates, well preserved on the right side, are large and extend from the maxillary tuberosities to the anteromedial edge of the glenoid fossa, as in Turkanapithecus. They are similar in size and shape to those observed in Symphalangus specimens at a similar stage of dental development, but are relatively larger than the lateral pterygoid plates of other extant hominoids, either immature or adult.

Dental morphology and development

The permanent dentition is unerupted and includes complete I1 and M1 crowns, along with partly formed crowns of I2, C–P4, and nearly complete M2 crowns (Fig. 2 and Supplementary Data 1). The morphology of the M1 matches that described for the genus Nyanzapithecus13,14,15, but differs from previously described species of this genus as noted in the diagnosis. The molars are relatively high crowned, and both M1and M2 display moderate crown waisting. The M2 is larger than M1, possesses a more prominent preprotocrista and prehypocrista compared with M1, and the M2 hypocone is positioned more distally relative to the other cusps than in M1. Both M1 and M2 have a crown diameter that is greater mesiodistally than buccolingually, and have the distinctive rhomboidal occlusal outline characteristic of other nyanzapithecines (Nyanzapithecus, Rangwapithecus,Turkanapithecus), Oreopithecus, Samburupithecus, as well as some extant hominoid specimens (Fig. 3b). In fact, the M1 shape falls exclusively within the nyanzapithecine range among fossil taxa and at the upper range of Symphalangus among extant taxa. The occlusal surfaces of the molars show conical, crowded, but well-defined cusps, combined with a relatively small, restricted trigon, and a distally offset hypocone. A prominent mesial cingulum is present, which continues around the protocone as a lingual cingulum, terminating at the hypocone. A short buccal cingulum is also present between the paracone and metacone. The occlusal surface enamel in KNM-NP 59050 is relatively smooth, with muted cuspal ridging and without the extensive wrinkling and cresting seen in Rangwapithecus.

The central incisor of KNM-NP 59050 is stout and the crown is spatulate, with the incisal–cervical long axis of the crown being canted mesially with respect to the long axis of the root. The lingual surface bears a strong cingulum that is angled incisally from distal to mesial, and prominent mesial and distal marginal ridges are present. The cingulum is continuous with the distal marginal ridge, but is separated from the mesial marginal ridge by a shallow cleft, creating a rectangular, arched fovea, with a distinct enamel bulb or tubercle at its superior-lingual edge. The mesiodistal length of the crown is short relative to the length of M1 (Fig. 3c), which is a distinctive feature found in Nyanzapithecus and hylobatids (Symphalangus, particularly). Overall, I1 is very similar in shape to the I1 of KNM-MB 11842, a fragmentary premaxilla/maxilla assigned to N. pickfordi13 (Extended Data Table 1), and Bar 217′02, an isolated upper left I1 assigned to N. cf. pickfordi16.

Dental development was analysed using synchrotron virtual histology17,18 (Extended Data Fig. 6). KNM-NP 59050 presents a long-period line periodicity of 5 days (Extended Data Fig. 6c), following expectations for a primate of this size19. The neonatal line was identified in the M1s and the I1s, and a developmental sequence for the permanent teeth could be built using stress lines in dentine to match teeth (Fig. 4 and Extended Data Fig. 6). An age at death of 485 ± 40 days was established. The two-dimensional relative enamel thickness index20 of the M1s, measured in the mesial developmental plane21, is 12.2, which is intermediate between Symphalangus (10.8) and Hylobates (13.3) (ref. 22). KNM-NP 59050 shows an unusually advanced I1, developing at a similar speed to the M1. Among catarrhines, this pattern is found only inHylobates23 and Hoolock but not in Symphalangus23 or Nomascus.

Figure 4: Dental development of KNM-NP 59050.
figure 4

On the basis of virtual histological slices (Extended Data Fig. 6), the crown (green) and root (purple) development of the upper right dentition is plotted from prenatal initiation of the M1 and I1 to death at 485 days after birth. In addition to birth (magenta) and death (black), the coloured lines represent reference stress lines in the dentine shown in the slices (Extended Data Fig. 6c) at the following day counts: blue, 265; green, 330; cyan, 365; yellow, 420; and red, 455. Sides of teeth indicated by B, buccal; D, distal; L, lingual; M, mesial.

PowerPoint slide

Full size image

Endocranial volume, olfactory fossa, and inner ear

A preliminary reconstruction of the cranium indicates an endocranial volume of 101 ml, which is projected to be close to the adult value if KNM-NP 59050 followed the correlation between brain growth and dental development seen in extant hylobatids (Extended Data Table 1b). Scaled against body mass (Supplementary Note 1), this endocranial volume is smaller than seen in hylobatids, and close to values obtained for Turkanapithecus and Oreopithecus (Extended Data Fig. 7a). The olfactory fossa is shallow and underneath the frontal lobes (Fig. 1h), as seen in extant catarrhines and unlike the larger and rostrally projecting fossae in Aegyptopithecus, Saadanius, Victoriapithecus24, and Afropithecus. The bony labyrinth of the inner ear is preserved on both sides. In primates, this structure fully matures before birth, and in KNM-NP 59050 the overall size is closer to that of hylobatids than to hominids (Extended Data Fig. 7c–n), reflecting similarities in cranial size. However, in shape the labyrinth of KNM-NP 59050 uniquely shares with extant great apes a distinctly low-arced anterior semicircular canal25 (Extended Data Fig. 7c–f and Supplementary Table 1). The lateral semicircular canal is low-arced as well, as seen in Pan andPongo25 among extant hominoids, and inOreopithecus and Aegyptopithecus among extinct catarrhines. The arc size of the lateral canal is small relative to those of the anterior and posterior ones, a feature observed in Aegyptopithecus, Saadanius, Oreopithecus, and Rudapithecus, but not in extant hominoids (Supplementary Table 1). Scaled against body mass, the semicircular canals are relatively small, as in extant great apes, Saadanius,Rudapithecus, and Hispanopithecus, and unlike the large canals of hylobatids26 (Extended Data Fig. 7b).

In summary, in addition to the dental features given in the species diagnosis, the cranium of N. alesi can be characterized by the following, ontogenetically stable morphology: spatulate incisors, enamel with a relative thickness index of 12.2, a relatively small snout, wide interorbital distance, protruding and bar-like inferior and lateral margins of the orbit, narrow nasal aperture, shallow maxillae, a zygomatic root that originates low on the maxilla, large pterygoid plates, external acoustic meatus represented by a fully ossified ectotympanic tube, shallow and non-projecting olfactory fossae, and semicircular canals that are small-arced relative to body mass. Several of these features are also shared among other related genera for which partial crania are known (for example, Oreopithecus and Turkanapithecus).

Phylogenetic analysis

The attribution of KNM-NP 59050 to Nyanzapithecus provides an opportunity to clarify the phylogenetic relationships of this relatively rare genus, on the basis of a much more comprehensive character evaluation than has been previously possible8,14,15,27. A cladistic analysis places N. alesi firmly within the nyanzapithecines, along with Rangwapithecus, Turkanapithecus,Oreopithecus, and Rukwapithecus; these genera, along with afropithecines, form part of a sister clade to the crown hominoids (Fig. 5 and Extended Data Fig. 8) rather than being stem catarrhines (contrary to ref. 15). A close relationship between Oreopithecus andNyanzapithecus has been suggested previously on the basis of dental morphology13. Our analysis, which scores ontogenetically stable cranial characters for N. alesi, supports this hypothesis, in contrast to studies that place Oreopithecus among the crown hominoids28,29,30,31.

Figure 5: Phylogenetic placement of N. alesi.
figure 5

Strict consensus of the ten most parsimonious trees from the unscaled phylogenetic analysis of 265 cranial and postcranial characters (tree length = 1383; consistency index = 0.289; homoplasy index = 0.711; retention index = 0.597). N. alesi is placed within the nyanzapithecines, which with the afropithecines form the sister group to crown hominoids. See also Extended Data Fig. 8 for bootstrap values and results of scaled analysis.

PowerPoint slide

Full size image

Discussion

KNM-NP 59050 is the first nearly complete African hominoid cranium recovered from between 17 and 7 Myr ago, and, to our knowledge, the most complete Miocene ape cranium yet described. KNM-NP 59050 provides critical evidence about the cranial anatomy of nyanzapithecines and, more broadly, of hominoids during an under-sampled time period in the African Miocene. Phylogenetic, cranial, and dental analyses of KNM-NP 50950 offer compelling insight into the ancestral morphology that gave rise to the clade containing extant apes and humans. N. alesi is similar to some hylobatids in aspects of overall cranial morphology and dental development, including enamel thickness, advanced I1 crown formation, and M1 crown formation time23. However, N. alesi is distinctly different from hylobatids in the relatively small size of its semicircular canals. This property is functionally relevant for the perception of motion, and suggests that N. alesi would have exhibited a slower, less agile mode of locomotion than the acrobatic brachiation of extant hylobatids26. This finding agrees with a previous study of a partial humerus provisionally attributed to Nyanzapithecus32. Some of the cranial similarities with extant hylobatids, including a relatively short face, a broad interorbital distance, and orbits with projecting inferior rims, are present not only in N. alesi, but also in other proposed nyanzapithecines, such as Turkanapithecus and Oreopithecus, as well as pliopithecoids, dendropithecoids, and colobines to varying degrees. This finding emphasizes the fact that multiple clades evolved hylobatid-like craniofacial morphology in parallel during catarrhine evolution. Hence, this general phenotype cannot be taken as evidence of shared ancestry with extant hylobatids in the absence of more convincing synapomorphies. As a case in point, our phylogenetic analysis suggests that the recently described Pliobates cataloniae, considered to be closely related to crown hominoids in part on the basis of hylobatid craniofacial similarities12, is more likely to be a member of the stem catarrhine group Pliopithecoidea (Fig. 5).

Since its initial description over 30 years ago13, the catarrhine genus Nyanzapithecus has been known mostly from isolated dental specimens. KNM-NP 59050 represents the first substantial cranial material of Nyanzaptithecus, and it confirms that the genus, and nyanzapithecines more broadly, possess a fully ossified tubular ectotympanic, a derived feature linking the group with crown catarrhines relative to more primitive taxa such as pliopithecoids. KNM-NP 59050 is currently the only described specimen of the new species N. alesi. However, similarities in the size and shape of the M1 shared between N. alesi and isolated teeth previously referred to Nyanzapithecus sp.33 or N. cf.pickfordi16 from Kipsaramon, Kenya (approximately 15.83–15.36 Myr (refs 33, 34, 35)) suggest that these specimens may also be part of the N. alesi hypodigm. Nyanzapithecines were a long-lived and diverse group of Miocene hominoids that are probably close to the origin of crown hominoids. They first appear in the fossil record during the latest Oligocene of Africa27, and persisted until perhaps the late Miocene, if the enigmatic Eurasian species Oreopithecus bamboli is indeed a late-surviving member of this clade13,14,15,28.

Methods

Locality description and specimen discovery

Napudet (2° 58.103′ N, 35° 51.969′ E) is located in South Turkana, Turkana Basin, Kenya (see Extended Data Fig. 1g). Sedimentary strata at Napudet were identified as part of the regional mapping of the area by the Kenya Geological Survey36. The Koobi Fora Research Project discovered the locality in 1990. I.N. relocated the site in 2013, with the help of a team from the Turkana Basin Institute, and directed surveys in 2014 and 2015. KNM-NP 59050 was discovered in 2014 by team member J. Ekusi, and prepared by C. Kiarie at the Turkana Basin Institute.

Stratigraphy and age

The Napudet Hills are formed by footwall uplift resulting from movement along the Napudet–Loperot Fault, with Miocene and Pliocene strata elevated above the surrounding low relief plains of the South Turkana Desert36,37 (Extended Data Fig. 1g). At the northern end of the Napudet Hills, sedimentary strata comprise four sequences, the lowest of which is a volcaniclastic interval informally termed the Emunyan beds here (Extended Data Fig. 2a and Supplementary Table 2), containing abundant fossil wood and scattered but well-preserved vertebrate fossils. The hominoid fossil KNM-NP 59050 was a surface find at the edge of exposures of the prominent brown bedded tuffs in the lower part of the Emunyan beds. Matrix sediment removed during preparation of the cranium was consistent with derivation from this sedimentary unit.

A basalt flow that underlies the fossiliferous Emunyan sedimentary sequence is dated here by Ar–Ar to 13.31 ± 0.04 Myr. Preliminary alternating field demagnetization of an oriented sample of this basalt yielded a normal geomagnetic polarity, suggesting that the basalt formed during either C5AAn (13.03–13.18 Myr) or C5ABn (13.36–13.61 Myr)38. Magnetic polarity stratigraphy of the overlying Emunyan beds demonstrated normal polarity for the level of the hominoid fossil, and a normal-to-reversed transition approximately 5 m above the level of the fossil. Assignment of this transition to the top of Chron C5AAn would suggest an age of slightly older than 13.0 Myr, while correlation to the top of C5ABn would imply a slightly older age of about 13.4 Myr for the hominoid.

The 40Ar–39Ar age of sample 15-NPD-03 was determined by high-precision step-heating technique on a MAP-215-50 mass spectrometer in the Noble Gas Laboratory at Rutgers University (Extended Data Fig. 2b and Supplementary Data 3).The sample was petrographically evaluated, crushed, sieved to 300–600 μm size range, washed in distilled water in an ultrasonic bath, and dried in an oven at ~80 °C. The sample was hand-picked and loaded into aluminium irradiation disks along with multiple splits of monitor minerals and wrapped in aluminium foil, and neutron irradiated at 1,000 kW for 20 min using cadmium-foil shielding in the central thimble facility of the US Geological Survey TRIGA reactor in Denver, Colorado. A 40-W CO2 laser with a jogging square laser beam (6 mm × 6 mm) was used as the thermal source for the incremental-heating experiments. The irradiation parameter J was determined by multiple total-fusion analyses of co-irradiated monitor mineral Fish Canyon Sanidine (FC-2 = 28.201 ± 0.046 Myr)39. Age calculations were made using currently accepted decay constants and isotopic abundances40 (1977): λε = 5.81 × 10−11 yr−1, λβ = 4.962 × 10−10 yr−1, 40K/Ktotal = 1.167 × 10−4. The following interfering neutron reaction from Ca and K values were used39,40: (36Ar–37Ar)Ca = 2.64 ± 0.02 × 10−4; (39Ar–37Ar)Ca = 6.73 ± 0.04 × 10−4; (38Ar–39Ar)K = 1.34 ± 0.02 × 10−2 from refs 41, 42, and (40Ar–39Ar)K = 2.85 ± 0.5 × 10−4 correction determined from measurements of kalsilite glass. When we plotted the incremental release spectra of the plateau steps identified for sample 15-NPD-03 on an isochron plot, it was possible to observe that the 40Ar–36Ar intercept was not purely atmospheric. Thus, we corrected the data using these trapped 40Ar–36Ar components. The corrected plateau 40Ar–39Ar age obtained, 13.31 ± 0.04 Myr, is considered to be the most representative for this sample (Extended Data Fig. 2b).

X-ray microtomography

KNM-NP 59050 was scanned using propagation phase-contrast X-ray synchrotron microtomography at beamline ID19 of the European Synchrotron Radiation Facility in Grenoble, France. The purpose was to visualize the specimen from a full overview down to virtual histology for the study of dental development. Four configurations were therefore used, providing voxel sizes of 28.06, 12.86, 3.44, and 0.74 μm. All acquisition parameters are summarized in Supplementary Table 3. Extant hominoid crania were scanned for comparative purposes using beamline BM05 of the European Synchrotron Radiation Facility in polychromatic mode (average energy between 100 and 130 keV), the GE phoenix v|tome|x s240 at the American Museum of Natural History, New York, and the BIR ACTIS 225/300 of the Max Planck Institute for Evolutionary Anthropology, Leipzig. Voxel sizes varied between 22.93 and 53.19 μm depending on the size of the specimens. VGStudioMax 3.0 (Volume Graphics), Avizo 7.1 (FEI), and Amira 5.6 (FEI) were used for two- and three-dimensional visualization, segmentation, reconstruction, and measurements.

Preliminary reconstruction of the cranium

To make meaningful comparisons possible, a preliminary retrodistortion of KNM-NP 59050 was attempted. This was done first on six orthogonal views using two-dimensional distortion maps to restore symmetry and to compensate for the major fractures (Extended Data Figs 1a–f and 3a). The two-dimensional distortion maps were then applied sequentially in three dimensions to the original image volume of KNM-NP 59050 (Extended Data Fig. 3b). This approach of correcting plastic deformations and fractures should be reasonably reliable for large-scale aspects and to obtain a preliminary estimate of the endocranial volume, but not necessarily with respect to more detailed morphology. A full reconstruction of the cranium will require extensive segmentation of all the bony components affected by cracks and plastic deformation, and subsequent three-dimensional correction of any distortions and misalignments.

Dental development and synchrotron virtual histology

Dental development in N. alesi was investigated quantitatively, using virtual synchrotron palaeohistology17 to examine the incremental lines preserved in the enamel and dentine.

Long-period lines and stress patterns were visible in scans with a voxel size of 3.44 μm of complete teeth of KNM-NP 59050, and long-period line periodicity could be seen in scans with a voxel size of 0.72 μm.

Diagenetic factors made it difficult to observe the enamel incremental lines in many of the dental germs. Only the right I2 exhibited a good enough contrast of the microstructures to establish a long-period line periodicity of 5 days (Extended Data Fig. 6d), and this figure was used for the rest of the analysis to calculate time in days.

Although enamel microstructures were not widely visible across the dentition in the 3.44 μm scans, dentine showed a very clear incremental pattern at this resolution, with good visibility of Andresen lines (equivalent to Retzius lines in enamel), especially in the right I2 (Extended Data Fig. 6c). Andresen lines were also visible in many other teeth, but not at a clarity that would allow the precise timing of tooth initiation to be documented. Nonetheless, all elements of the dentition exhibit a fairly clear stress pattern with recognizable accentuated lines. By comparing the pattern of the right I2, calibrated against time using the Andresen line counts and periodicity, and with identification of the same stress lines in other teeth (Extended Data Fig. 6b, c), it was possible to retrieve the relative dental development sequence for the whole dentition.

For developing tooth germs, the dentine daily secretion rate along the cusp axis (from dentine horn to developing pulp cavity) appears to be relatively constant, as long as the lateral enamel is still extending. We used this property to estimate dental initiation for all the teeth, by measuring the distance along the developmental axis of the cusps, between the dentine horn and the first recognizable reference stress line. Dentine axial daily secretion rates were calibrated for the right I2 and the left M1 and then applied to all the other teeth.

The first visible accentuated line in the M1s was interpreted as the neonatal line, and this line was observable in all cusps except the hypocone. After careful superimposition of the stress pattern across the whole dentition, it was clear that the neonatal line was also visible in the I1s as the uppermost stress line, indicating a prenatal initiation not only of the M1s, but also of the I1s.

By combining these data, it was possible to quantify the complete dental development (Supplementary Data 4), and to build developmental charts (Fig. 4 and Supplementary Data 4). The age at death was then measured as 485 days after birth with an error margin of 40 days. Crown formation times of the M1s (paracone) and of the I1s are 1.05 and 1.21 years, respectively (average of both sides).

Of special interest is that the I1s inN. alesi initiate development very early and complete their growth at the same time as the M1s. Considering the rapid extension of the I1 roots, which are longer than those of the M1s, the I1s would probably have erupted before, or at the same time as, the M1s. This proposition is strengthened by the advanced root resorption of the deciduous I1s and I2s, suggesting ongoing eruption of the permanent I1s, followed by the I2s.

The general dental development pattern of KNM-NP 59050, and the advanced I1 development in particular, were studied in more detail by making comparisons with extant juvenile hominoids and cercopithecoids. These included Pan troglodytes (10), Gorilla gorilla (3), Pongo pygmaeus (4), Homo sapiens (6), Hoolock sp. (4),Hylobates muelleri (1), Nomascus hainanus (1); and the cercopithecoidsPapio ursinus (1), Cercopithecus petaurista (1), Macaca sp. (2), and Macaca nigra (1). These specimens are in the collections of the Musée des Confluences de Lyon and were scanned at the European Synchrotron Radiation Facility, except for theHoolock material, which is housed in, and was scanned at, the American Museum of Natural History (New York), as well asHylobates and Nomascus specimens, which are housed at the Museum für Naturkunde, Berlin, and were scanned at the Max Planck Institute of Evolutionary Anthropology (Leipzig). Results of these comparisons show that the unusual pattern of advanced development of the I1s is found only in Hylobates and Hoolock.

Three dental development characters were coded as part of the character matrix used in the phylogenetic analyses (Supplementary Data 5). These were the relative developmental timing of I1 versus M1, M1 versus M2, and M2 versus P4. These data were taken preferentially from upper teeth when available, but also from lower ones where necessary (assuming a similar developmental pattern in upper and lower dentition). The character states were obtained by direct observations (including online tomography database) and from the literature, on the basis of computed tomography of juveniles, documented eruption patterns, or histological data.

Semicircular canals of the inner ear

The left bony labyrinth of KNM-NP 59050 is best preserved and analysed here. It was compared quantitatively on the basis of arc shape and size of the semicircular canals, using features known to distinguish hominids, hylobatids, and other catarrhines25,26,43,44. Data are provided in Supplementary Table 1. For fossils and the extant hominoids, the arc height and width of each canal25 were measured from three-dimensional surface models extracted from microtomographic scans (Extended Data Fig. 7a–l). In the case of extant hominoids, this concerned a model representing the mean shape and size of the sample. Data were also obtained for additional anthropoid species43,44, for use in the phylogenetic analyses. The mean radius of curvature of the semicircular canals was scaled against body mass (Extended Data Fig. 7h, ref. 45 and Supplementary Note 1).

Morphological and cladistic analyses

Specimens examined for the morphological and phylogenetic analyses derived from collections at the American Museum of Natural History, the Harvard Museum of Comparative Zoology, the National Museums of Kenya, Nairobi, and the Center for the Study of Human Origins at New York University. Hominins were not included in the comparative sample, and not considered here when using the terms ‘hominoid’ and ‘hominid’. Extant juvenile specimens were selected to encompass the dental age of KNM-NP 59050, from erupted deciduous dentition only to full eruption of the M1 and/or I1, some with M2 starting eruption as well (Supplementary Data 2). Adult specimens were defined as those with a completely erupted permanent dentition.

Observations and measurements are based on original specimens, as well as on high-quality casts, published photographs, and data provided in the literature. All juvenile data were collected by the authors; adult data were collected by the authors whenever possible, with additional data from ref. 46, PRIMO (access courtesy of E. Delson), and the literature (Supplementary Data 2). Whenever possible, individual specimens were used for statistical comparisons, but in a few cases published species averages were used to include key taxa. Adult features were assessed for KNM-NP 59050 through quantitative and qualitative comparisons with the above-referenced sample of extant juvenile and adult hominoids. Any given morphological feature was deemed to hold constant throughout ontogeny if present in both adult and juvenile extant ape specimens of the taxa examined41. These ontogenetically stable features are discussed throughout the main text.

Comparative analyses were based on standard craniometric and dental measurements (Extended Data Table 1 and Supplementary Data 2). Those of KNM-NP 59050 were taken from the original specimen as well as from a high-resolution three-dimensional surface visualization derived from the microtomographic images (Extended Data Figs 1a–f and 3a, b). For comparisons of juvenile crania, the measurements were size-adjusted by dividing each by the geometric mean of 13 cranial measurements (Supplementary Data 2). For comparisons of adult crania, the measurements were size-adjusted on the basis of the square root of M1 area to enable comparisons with key fossil specimens that are invariably too fragmentary to obtain a geometric mean of an adequate set of cranial measurements. The two methods of size adjustment are broadly similar, given that the geometric mean used here and the square root of M1 area are highly correlated (r = 0.935; r2 = 0.874; P < 0.001; 61 adult extant ape specimens). Nevertheless, juvenile and adult specimens were not directly compared, as we only focused on the pattern of morphological differences between taxa. In particular, we examined whether specific patterns were consistently maintained from infants to adults to assess the ontogenetic stability of particular features. For KNM-RU 7290, both the actual M1 area and a reduced estimated area were used, aimed at correcting for the megadontia exhibited by E. heseloni47 (Supplementary Note 1). Statistical analyses were performed in PAST 3.14 (ref. 48) and SPSS version 22.0.

Phylogenetic analyses

In addition to general qualitative and quantitative anatomical comparisons, we conducted a parsimony analysis of morphological characters to assess the phylogenetic position and likely evolutionary relationships ofN. alesi. The morphological character matrix of ref. 8 was supplemented with cranial characters from ref. 29 and several other sources12,49,50 (Supplementary Data 5). Twelve new characters were also included, and several character states were modified to accommodate polymorphisms or features found in the broadened taxonomic sample. In some cases, scores differed from the datasets from which they were derived on the basis of our independent qualitative and quantitative assessments of morphology (as indicated in Supplementary Data 5). The taxon sample of ref. 8 was expanded to include additional fossil catarrhines, including Saadanius, Rukwapithecus,Oreopithecus, Afropithecus, Kenyapithecus,Ouranopithecus, Sivapithecus, Lufengpithecus,Hispanopithecus, Pierolapithecus, and the recently described Pliobates12. We did not include Mabokopithecus in our analyses because the hypodigm and its relationship to Nyanzapithecus are unresolved and under study51. For these additional taxa, scores were based on our own data and observations as well as information provided in the literature; many character codings for Saadanius andRukwapithecus were taken from ref. 27, while codings for Pierolapithecus, Hispanopithecus, and Pliobates (with modifications) were taken from ref. 12. All characters and character states are provided in Supplementary Data 5 and the matrix is provided as a Nexus file in Supplementary Data 6. In total, 265 characters scored for 47 taxa were included in the analysis, incorporating characters of the skull and postcranial skeleton.

Characters were considered ordered whenever it could be assumed that a population probably passed through an intermediate state to get to an extreme state on either side. In these cases, ordering characters is a much more faithful representation of the evolutionary process (for example, a population does not typically evolve from a small body mass to a large body mass without passing through an intermediate body mass through directional selection)49. Polymorphisms were coded as intermediate states between two fixed states whenever possible; simulations suggest that this coding system increases the accuracy of the resulting trees52. All other characters were left unordered. Two analyses were performed: one with characters unscaled (that is, all steps given equal weight), and one with multistate, polymorphic characters scaled so that these characters had the same weight as binary and unordered characters (that is, polymorphic steps were downweighted relative to steps between fixed states) (Fig. 5, Extended Data Fig. 8 and Extended Data Table 2). Adult character states were estimated for KNM-NP 59050 through quantitative and qualitative morphological comparisons with the above-referenced sample of extant infant and adult hominoids (see, for example, Extended Data Figs 4 and 5). A character state was deemed to hold constant throughout ontogeny if it was present in both adult and infant specimens close to the same dental age as KNM-NP 59050. Wherever it could be safely assumed that a character state held through ontogeny, we scored KNM-NP 59050 accordingly53.

The resulting matrix was analysed using a 10,000 replication, random addition sequence heuristic search in PAUP 4.10b. To provide an estimate of clade support, a 1,000-replication bootstrap procedure with replacement was performed. A sample of platyrrhines (Aotus (Cebus, Saimiri)) along with the primitive catarrhines Catopithecus and Aegyptopithecus were assigned and constrained as successive outgroups, with the ingroup composed of Saadanius, pliopithecoids, Old World monkeys, dendropithecids, proconsulids, Pliobates, hylobatids, fossil hominids (Sivapithecus, Kenyapithecus, Pierolapithecus,Hispanopithecus, Ouranopithecus, Lufengpithecus, and Oreopithecus), and extant hominids (Pongo, Gorilla, and Pan). Broad-level taxonomy follows ref. 1.

Data availability

KNM-NP 59050 is available for study at the National Museums of Kenya, and data analysed here are provided in the article and its Supplementary Information. Image datasets of comparative specimens can be accessed at http://paleo.esrf.eu, http://africanfossils.org, and http://morphosource.org. The species is registered in ZooBank (http://zoobank.org; LSID urn:lsid:zoobank.org:act:BE0A6575-AD3A-4415-A169-6ABC6B8280E2).